Your questions, our answers
Everything you want to know about Smart-Portal.ai and Business Virtual Assistants — before you even reach out.
What is an MES?
An MES (Manufacturing Execution System) is a software system that monitors and controls production operations in real time. It bridges planning (ERP) and the factory, ensuring traceability, performance, and quality.
Read more
Where the ERP plans and manages the "how much" and the "by when," the MES manages the "how" and the "with what" in the factory: it records every manufacturing step — raw materials consumed, operators involved, machines used, machine parameters, quality checks — as it happens. It's this real-time factory ↔ information system loop that sets an MES apart from a simple production management tool.
More details on the MES FAQ from the publisher isiTecc (in French)
Why is the MES the ideal candidate for agentic AI?
Because the MES is the only system in the company that acquires, enriches, and stores, in a unique and structured way, the full set of production data: manufacturing orders, real-time per-operation tracking, quality declarations, machine deviations, material traceability, operator events, line performance.
The ERP only sees the theoretical plan. The CMMS only sees maintenance. IoT sensors push raw streams, with no business context. Only the MES reconciles all these signals into a coherent, timestamped, attributed data model — exactly what an agentic AI needs to reason without hallucinating.
That's why Smart-Portal.ai relies on the MES as its foundation: each Business Virtual Assistant (BVA) works on data that is already cleaned, normalized, and contextualized. No data pipeline tinkering — the AI's added value goes straight to searching, proposing solutions, and even applying them.
What is a Business Virtual Assistant (BVA)?
A BVA is an AI agent specialized in one industrial domain: production planning, quality, maintenance, manufacturing engineering, or factory operations. Each BVA continuously digests your production data, cross-references information across domains, and proposes quantified decisions before problems become visible.
Smart-Portal.ai ships 8 interconnected BVA agents: 5 for the domains (production planning, quality, maintenance, manufacturing engineering, production) and 3 for decision-makers (production management, plant management, IT). All run in parallel, 24/7, and work together — not in silos.
Unlike a classic alert that flags a problem already happening, a BVA anticipates from weak signals and arrives with quantified scenarios.
How does Smart-Portal.ai integrate with my existing MES?
Smart-Portal.ai connects to your existing data sources — MES, ERP, CMMS, IoT sensors, Excel schedules — with no migration and no rebuild. Your teams keep their tools: BVA agents read data where it already lives, in real time.
The recommended path is MCP (Model Context Protocol), the interoperability standard built for AI agents. MCP lets each BVA access your MES natively, securely, and in both directions — fastest to deploy, most durable, and our recommendation as soon as your MES supports it.
If your MES doesn't expose an MCP interface yet, integration is done through standard connectors (OPC-UA, REST, SQL databases, flat files) and adapters built for the leading MES platforms. Connection typically takes a few days to a few weeks depending on the maturity of your information system.
Smart-Portal.ai is published by isiTecc, MES specialist since 2016 — you keep your MES, you add agentic intelligence on top.
Where is my data stored and how is it secured?
Your data stays yours. Smart-Portal.ai can be deployed in a sovereign European cloud, hosted by isiTecc (OVH hosting, Roubaix) on an instance dedicated to your company; in your private cloud; or on-premises, depending on your sovereignty requirements.
No customer data is used to train AI models — your industrial know-how never leaves your perimeter. Communications are end-to-end encrypted, and every action is traced per user. Smart-Portal.ai complies with GDPR (EU Regulation 2016/679), and its architecture is built to meet ISO 27001 and NIS2 requirements: automatic traceability of BVA decisions makes it easier to produce incident reports within the timelines NIS2 imposes (24 hours for the initial alert, 72 hours for full notification).
The IT BVA, dedicated to your CIO, monitors the architecture continuously, detects behavior anomalies, and blocks deviations automatically.
How long does deployment take, and what ROI should I expect?
Deployment happens in three phases: data connection (a few weeks), progressive activation of BVA agents by domain (1 to 2 months for the first one, faster for the next), and autonomy ramp-up at your own pace — supervised, then hybrid, then autonomous.
You don't wait until the end of the project to see benefits: each BVA delivers value from the moment it's enabled.
ROI depends on your industrial maturity, your volume, and the criticality of your domains. Our team helps you size it against your situation.
What's the difference between Smart-Portal.ai and a classic MES?
A classic MES records, tracks, and reports what happened in the factory. It gives you late, fragmented indicators that recommend nothing.
Smart-Portal.ai changes the nature of the tool: it interprets, anticipates, and proposes. Where an MES tells you what happened, Smart-Portal.ai tells you what's about to happen, why, and what to do — with quantified scenarios.
This is what Industry 5.0 calls "human in the loop": AI proposes, humans decide. You don't replace your MES — you add an agentic intelligence layer that digests, cross-references, and interprets what the human brain can no longer process in time.
How much autonomy can I delegate to a BVA?
You set the cursor, domain by domain, action by action. Three levels are available.
Supervised — the BVA detects and alerts; you decide. Ideal to start, or on sensitive production runs.
Hybrid — the BVA proposes several quantified scenarios; you approve in one click. A good fit for most situations.
Autonomous — the BVA acts within the boundaries you've defined. Every decision is logged and auditable.
The cursor goes up as confidence grows and can come back down at any time. Example: during a scheduler's vacation, switch to hybrid to support a temporary replacement; on a sensitive production run, switch back to supervised to take direct control.
It's chosen autonomy, not forced automation.